Find papers | Search over 125MM academic papers in Elicit

Elicit’s “Find papers” workflow lets researchers search across roughly 200 million academic papers using natural-language queries, then turn the most relevant results into question-specific summaries and structured data. Instead of relying on keyword matching or complex Boolean strings, the system uses semantic search to find papers with similar meaning—often pulling in relevant work even when the exact keywords don’t overlap. That matters because it speeds up early-stage literature discovery, when the hardest part is usually figuring out what to search for and how to narrow down millions of candidates.

A typical session starts on the Elicit homepage (elicit.com) and selects the “Find papers” workflow. Users can ask a descriptive question such as: “What are the long-term effects of invasive species in the Pacific Northwest?” Elicit then searches its database and returns a ranked list of the most relevant papers, using titles and abstracts as the primary signals. The interface shows quick, dynamically generated summaries of each paper’s abstract tailored to the user’s question, helping readers judge relevance without opening every result.

Results appear in a table where each row corresponds to a paper and includes metadata such as authors, journal, citation count, and links to the DOI and (when available) the PDF. Users can click through to view the paper and see the abstract alongside the DOI landing page. Elicit also supports manual sorting and filtering: papers can be ordered by recency or citation count, constrained to date ranges, and filtered to include only open-access PDFs when full text is needed for richer extraction.

To get beyond basic paper lists, Elicit’s “columns” feature adds structured fields extracted from the papers. Predefined columns—especially common in biomedicine—can be added with a click, and users can create custom columns by specifying exactly what information they want. For example, a user can request the invasive species studied in each paper; Elicit extracts the relevant entities from the abstract or full text (depending on access) and displays the results in the table. The system also provides traceability: users can click quotes to see where the extracted information came from and open the underlying paper to verify context.

When semantic relevance misses relationships users care about—or when a topic is narrow—Elicit supports citation network searching, including citation trail and citation chasing. After selecting a seed set of papers (for instance, those focused on crayfish), Elicit searches the citation graph for both references and future citations, surfacing additional related papers that connect through the scholarly network.

Cost and workflow tradeoffs are explicit. “Find papers” is described as the cheapest workflow in credits, but adding more columns and searching across more papers increases credit usage. For downstream work, users can export results to CSV and BibTeX for reference managers such as Zotero. The workflow is designed to scale from quick discovery to structured extraction, with higher-accuracy column options available later via subscription features and follow-up capabilities.